Assignment7 - solutions
BMI 713: Computational Statistics for Biomedical Sciences
Assignment 7
Simple Linear Regression
1. To study the relationship between a father's height and his son's height, Karl Pearson (1857-1936) collected the data of heights from 1078 father-son pairs.
(a). Get the dataset by the following R commands:
install.packages("UsingR") library(UsingR) data(father.son)
Then the data frame father.son contains the 1078 observations on 2 variables: fheight (father's height in inches, x) and sheight (adult son's height in inches, y).
(b). Draw a scatter plot of son's height versus father's height. Does the relationship appear linear? Sol'n. From the scatter plot below, we can see that the son's height tends to increase as the father's height increases.
> plot(fheight, sheight, xlab="Father's height (in)", ylab="Son's height (in)", xlim=c(58,78), ylim=c(58,80), bty="l", pch=20)
80
75
70
Son's height (in)
65
60
60
65
70
75
Father's height (in)
(c). Fit the simple linear regression of son's height on father's height. What are the estimated regression coefficients, a and b, respectively?
Sol'n. Denote the 1078 father-son pairs of observations as (x1, y1), ..., (xn, yn), where n = 1078. We will fit the linear regression model of son's height y on father's height x: y = + x + e, e ~ N(0, 2 )
Fit the linear model by the method of least squares, and the estimated regression coefficients are:
n
b
=
(xi
i =1
n
- x )(yi - (xi - x )2
y)
=
Cov(x, y) Var(x)
=
0.514
i =1
(slope),
and
a = y - bx = 33.89 (intercept).
We can also fit the linear regression model in R by the function lm:
> m summary(m) Call: lm(formula = sheight ~ fheight)
Residuals:
Min
1Q Median
-8.877151 -1.514415 -0.007896
3Q 1.628512
Max 8.968479
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 33.88660 1.83235 18.49 R.squared R.squared [1] 0.2513401
Here the R2 statistic is the proportion of the total response variation explained by the explanatory variable in the linear regression model.
(h). Draw a residual plot. Are the residuals normally distributed with constant variance?
Sol'n. The model assumptions of normal distribution and constant variance seem valid based on the residual plot below.
Residual Plot
5
0
Residuals
-5
64
66
68
70
72
Fitted values
(i). What are the estimated means of son's height given that his father's height is 72, 75, 60, and 63 inches, respectively?
(Notice that sons of tall fathers tended to be tall, but on average not as tall as their fathers. Similarly, sons of short fathers tended to be short, but on average not as short as their fathers. This phenomenon was first described by Sir Francis Galton, as "regression towards mediocrity", where the term regression came from. The regression effect ? phenomenon of regression toward the mean ? appears in any test-retest situation.)
Sol'n. The estimated means of son's heights are 70.9, 72.4, 64.7, 66.3 inches, given that his father's height is 72, 75, 60, and 63 inches, respectively.
(j). Given a father's height, we can use simulation method to construct the 100(1-)% confidence interval for the mean of his son's height. First draw 1000 samples each of size 1078 with replacement from the 1078 pairs of father-son heights, then from each sample fit a linear regression model by the method of least squares, and compute the estimated mean of son's height.
What are the mean and standard deviation of these 1000 simulated values?
Sort these 1000 estimated means in ascending order. Denote the 25th largest as h25 and the 975th largest as h975 , which are our estimates of the 0.025 and 0.975 quantiles of the sampling distribution for the mean of son's height. Then the 100(1-)% confidence interval for the mean
of the son's height is ( h25, h975 ). Compute the 95% confidence interval for the mean of son's height if his father is 72 inches tall.
Sol'n. Use the following loop to run 1000 simulations in R:
> n h.father h.son for (i in 1:1000) {
+
v ................
................
In order to avoid copyright disputes, this page is only a partial summary.
To fulfill the demand for quickly locating and searching documents.
It is intelligent file search solution for home and business.
Related download
- a predicted average peak expiratory flow fornormal
- d µ Ç u u î õ u î ì î ì î w ð õ wd
- x ap statistics solutions to packet 2
- free stall dimensions
- math 263 section 005 class 2 normal distribution and z
- height conversion occupational health
- lab 7 t tests
- body mass index table
- frequency distribution of height measurements
- assignment7 solutions
Related searches
- medical marijuana solutions state colleg
- medical marijuana solutions state college pa
- syneos health commercial solutions jobs
- syneos health commercial solutions locations
- come up with solutions synonym
- crm solutions providers
- onemain solutions disability form pdf
- advantage solutions training
- tmp advantage solutions training
- medical marijuana solutions pittsburgh
- one main solutions disability
- one main solutions find a form